Well it's definitely possible unless I'm a moron (correct me if I am). If you have the primes up to √n you can find all primes up to n because everything else will be composite of the primes you know already. This is basically what TFA is doing. What I was suggesting is that it would be cool if you could avoid defining rules explicitly for (3n+6), (5n+10), etc and have them appear from some feedback loop where we go from a sequence of primes up to n, to n^2, to n^4...

CSS4, let's do it.